The Prader-Willi syndrome (PWS) is caused by genomic alterations that inactivate imprinted, paternally expressed genes in human chromosome region 15q11-q13. IPW, a paternally expressed gene cloned from this region, is not expressed in individuals with PWS, and is thus a candidate for involvement in this disorder. The IPW transcript does not appear to encode a polypeptide, suggesting that it functions at the level of an RNA. We have now cloned a mouse gene, named Ipw, that has sequence similarity to a part of IPW and is located in the conserved homologous region of mouse chromosome 7. The Ipw cDNA also contains no long open reading frame, is alternatively spliced and contains multiple copies of a 147 bp repeat, arranged in a head-to-tail orientation, that are interrupted by the insertion of an intracisternal A particle sequence. Ipw is expressed predominantly in brain. In an interspecies (M.musculus*M.m.castaneus) F1 hybrid animal, expression of Ipw is limited to the paternal allele. We propose that Ipw is the murine homolog of IPW.
The IPW gene (1 ) is located within an imprinted region on human chromosome 15 that is deleted in 70% of cases of Prader-Willi syndrome (PWS) (2 ). Although the IPW transcript is widely expressed, spliced and polyadenylated, the 2.2 kb RNA sequence does not appear to have polypeptide-coding potential. The manifestations of PWS include mental retardation, neonatal hypotonia, hyperphagia leading to obesity, short stature and hypogonadism, suggesting a underlying hypothalamic defect. Since most non-deletion cases of PWS are due to uniparental maternal disomy for chromosome 15 (3 ,4 ) and are clinically indistinguishable from those with paternally derived deletions, the genes responsible for the phenotype must be imprinted, expressed only from the paternal allele and located in the deletion region.
IPW and two other paternally expressed genes have been identified within the deletion region, yet no causal connection has been made between the loss of expression of any particular gene and the manifestations of PWS. SNRPN codes for the small ribonucleoprotein-associated polypeptide N, a developmentally regulated protein component of the spliceosome expressed predominantly in neuronal tissues. It is paternally expressed in mice and humans (5 -7 ). ZNF127 (D15S9) encodes a zinc finger- containing protein, presumably a transcription factor, and is located in the proximal end of the deletion region (8 ). An allele-specific methylation difference has been documented for probes from this region (9 ), however, no expression studies have been published other than in abstract form (10 ).
A small number of PWS patients have deletions of <100 kb on the paternal chromosome that are located just centromeric of the SNRPN gene (11 ,12 ). These deletions result in loss of expression of imprinted genes in this region, including SNRPN (11 ) and IPW (13 ) and alter the differential methylation pattern of other loci, such as PW71B (14 ). For one family (PWS-U) lack of expression of ZNF127 has also been reported (15 ).
There is growing evidence that untranslated RNAs play a role in establishment or maintenance of a chromatin configuration that is necessary for the expression of nearby genes. The alternative hypothesis is that the transcription of untranslated RNAs merely reflects the transcriptional state of their chromosomal region, while the RNAs themselves have no function. Deletion in mice of the imprinted H19 gene, which encodes a non-translated RNA, disrupts imprinting in the region surrounding it (16 ). A disruption of DGCR5, a non-translated RNA from the DiGeorge syndrome region, is implicated in regional control of nearby genes (17 ). Finally, paternally expressed alternative SNRPN transcripts are proposed to be involved in imprint epigenotype switching. Non- translated exons upstream of SNRPN are disrupted in a subset of PWS and Angelman syndrome patients, whose disorder is apparently caused by a failure to reset the chromosome 15 imprint from one parental epigenotype to the other (18 ). In a further example, disruption of Xist, the inactive X-specific transcript, blocks X-inactivation in cis-configuration (19 ). In this last example, there is evidence that the Xist RNA may be directly involved in X chromosome inactivation by physical association between the Xist RNA and the chromosome (20 ).
IPW is located ~180 kb from the putative imprinting center required to maintain expression from this region on the paternal chromosome 15. The function of IPW is apparent neither from its expression pattern nor from its sequence. However, IPW may be involved in regional gene control in the PWS region, in a manner analogous to the proposed roles of other untranslated RNAs. In order to investigate further the function of IPW and its role in PWS and in gene regulation and chromatin configuration, we have searched for a mouse homolog. We report here the identification of an imprinted gene (Ipw) with sequence similarity to IPW that contains tandem 147 bp repeats and appears not to code for protein.
Seven overlapping clones (four [lambda] phage clones and three P1 clones) were identified by hybridization to a probe from human IPW exon 3 and by PCR screening of mouse genomic DNA libraries (Fig. 1 ). Partial sequencing of clone [lambda]mg1 revealed a 319 bp region of similarity to IPW with 79% nucleotide identity and four gaps (Fig. 2 ). The 5' end of this region (later named mouse exon C, 142 bp) is 70% similar to the 5' end of IPW exon 3, and the 3' end of mouse exon C corresponds to the middle of IPW exon 3. The 5' splice sites are at identical positions when the human exon 3 and mouse exon C are aligned. A splice site is present in mouse DNA at position 143 that is not present in human DNA. It is most unusual, however, that the similarity between the human and mouse sequences extends through the 3' splice site of mouse exon C into the adjacent intron for an additional 176 bp and diminishes thereafter.
Partial mouse brain cDNA clones were generated by three independent methods: rapid amplification of cDNA ends (RACE) (23 ) with primers derived from Ipw exon C, RT-PCR and cDNA library screening (Fig. 3 ). The RACE and RT-PCR cDNA clones contained overlapping sequences, termed exons A-G (Fig. 3 A). The 5' region of the Ipw cDNA contained multiple tandem repeats of a 147 bp sequence termed exon A. Two slightly different copies of exon A (A1 and A2) were found by sequencing subclones of [lambda]mg1. Exons A1 and A2 were present in phage [lambda]mg1 as single exons. The genomic structure of the other A repeats is unknown. Southern blot analysis with a probe from exon A revealed strongly hybridizing 0.8 and 2.5 kb EcoRI fragments in total mouse DNA and in DNA from the P1 clones. This finding suggests that the repeats are in separate exons, given the absence of restriction sites for EcoRI within each exon. It also suggests that the genomic sequence containing the A repeats is organized in tandem repeats of ~0.8 kb, defined by EcoRI restriction sites. It is noteworthy that the sequence of exon B bears some resemblance to that of exon A (Fig. 3 C), indicating an ancient evolutionary relationship.
A similarity search of sequence databases revealed no significant similarity of the sequence shown in Figure 3 C to known genes other than IPW. No sequence similarity between Ipw and human IPW was detected beyond the region noted in Figure 2 . The longest open reading frame (ORF), starting with a potential initiator methionine, predicts 104 amino acids and covers part of exon A2, all of exon A1 and part of exon B. No similarities were found when the protein databases were searched with this translated sequence. The potential initiator methionine is not surrounded by the consensus sequence (25 ). Furthermore, the program CODONPREFERENCE, which analyses sequences for coding potential using a codon usage table, detected no evidence for coding potential within the Ipw cDNA sequence. Most remarkably, however, the putative ORFs for IPW (1 ) and Ipw are not within the conserved region of sequence similarity and, thus, the conceptual translation products are completely different. We conclude from this comparative analysis that both genes are non-coding and that their function must be at the RNA level.
The expression of Ipw in mouse tissues was examined by Northern blot hybridization of total RNA with a probe from exons B and C. There was no signal in heart, liver or skin fibroblasts, but a strong smear ranging in size from 0.5 to 12 kb was seen in total brain RNA (data not shown). By using RT-PCR with primers amplifying exons B and C, a specific product was obtained only from brain RNA. Expression in other tissues was detected subsequently by using a hemi-nested RT-PCR, which detects very low abundance transcripts. This contrasts with human IPW, which is expressed at a low but relatively constant level in all tissues tested. Human IPW transcripts are easily detected by RT-PCR, without the use of the more sensitive hemi-nested procedure (1 ).
To determine the imprinting status of Ipw, we searched for sequence polymorphisms among different strains of mice by direct sequencing of PCR products from exons B and C, amplified using primers from flanking introns. No sequence polymorphisms were found in three strains of M.musculus (ICR outbred, C57BL/6J, 129/SvJ) and NIH3T3 cells. However, between these strains of M.musculus and M.m.castaneus a polymorphism was identified in exon B that consisted of `taaacc' in M.musculus and `taaccaccc' in M.m.castaneus.
This polymorphism resides outside of the potential ORF. To look for parent-of-origin specific expression, we examined the transcripts in tissues from an F1 product of a M.m.castaneus male crossed to a M.musculus female mouse. When hemi-nested RT-PCR products containing the exon B polymorphism were sequenced directly, only the paternal M.m.castaneus allele was expressed in all tissues (stomach, heart, testes, muscle, spleen, small intestine, brain, kidney, liver and large intestine) (Fig. 4 A). To ensure that the maternal allele was indeed present in the interspecies hybrid mouse, we performed PCR on liver DNA from the same animal with intron primers flanking exon B. Upon direct sequencing of this DNA PCR product, an overlapping signal from the two alleles was seen (Fig. 4 B). Parallel control experiments on RNA from a M.musculus mouse demonstrated that the M.musculus allele was expressed and amplifiable in this reaction (Fig. 4 C). We conclude that Ipw is an imprinted gene with expression limited to the paternal allele in all adult tissues examined, as is the case for IPW, its human counterpart.
Aside from the apparent lack of any other sequences homologous to IPW within the mouse genome, the specific evidence that Ipw is the murine homolog of IPW is 4-fold. The two genes share a region of 79% sequence similarity over 319 bp, although part of this region includes similarity between a human exon and a mouse intron. The Ipw gene is located in the central part of mouse chromosome 7, a region of known conserved synteny with human chromosome region 15q11-13, where IPW is located. Both genes lack protein-coding potential and both are imprinted, with expression limited to the paternal allele.
In contrast to IPW, which is ubiquitously expressed at a moderate level in all human tissues, Ipw is expressed primarily in brain, and elsewhere at a very low level. The presence of tandem repeats within the cDNA, encoded by separate exons, is a distinctive feature of Ipw. Restriction digests of genomic clones containing the exon A repeats suggest a repetitive organization of a larger genomic unit. As recently suggested, sequence repeats may be a common feature of imprinted genes (26 ). The high degree of sequence conservation between the copies of exon A may indicate a recent expansion of these repeats or, less likely, a functional importance of their sequences. In contrast to IPW, the mouse gene is expressed in multiple alternatively spliced forms. Whether they have distinct functions, either in different regions of the brain or during mouse development, remains to be investigated.
Ipw joins a growing list of genes known to be imprinted in the mouse. There are now nine imprinted genes on mouse chromosome 7, including three in the central region [Snrpn (22 ), Znf127 (10 ) and Ipw]. The other six genes are either in the proximal [Peg3, (27 )] or distal [Ins2, Igf2, H19, Mash2 and p57kip2 (28 ,29 )] region. One or more of these genes probably account for the imprinting effects observed in experimentally produced mice with uniparental disomies for regions of chromosome 7.
At least eight other mammalian genes produce spliced and polyadenylated transcripts with no long ORF and, therefore, appear to function at the level of RNA. Gene deletion experiments have been performed for two of these genes, H19 and Xist. Deletion of the imprinted H19 gene disrupts imprinting in the region surrounding it (16 ). H19 has been proposed to regulate the imprinting status of the neighboring genes Igf2 and Ins2 (16 ,30 ,31 ). This model supposes that the transcription of H19 blocks access of adjacent imprinted genes to shared enhancer elements, thus preventing their transcription. Lack of expression of H19 allows access to these enhancers, and transcription of the other genes proceeds.
The murine inactive X-specific transcript, Xist, contains no conserved ORF and plays a major role in X-inactivation (32 ). In support of this hypothesis, disruption of Xist blocks X-inactivation in the cis configuration (19 ), and transposition of a genomic fragment containing Xist and flanking sequences to autosomes causes them to be counted and chosen for the X-inactivation process (20 ). In the early mouse embryo, Xist is expressed exclusively from the paternal X chromosome (33 ).
Both H19 and XIST share >70% sequence conservation with their mouse homologs. Sequence comparison of additional mammalian homologs of XIST and H19 suggests that the sequence of these two untranslated RNAs is more conserved than introns of coding genes, but less conserved than coding exons. Furthermore, although the human and mouse XIST genes are similar in sequence and gene structure, only the 5' end of the XIST sequence appears to be conserved in other mammalian species (34 ).
Five other untranslated genes whose expression has not been assessed for imprinting have been described recently. The gadd7 cDNA was isolated from Chinese hamster cells as one of a group of genes induced by DNA damage. Overexpression of gadd7 leads to growth arrest in both hamster and human cells. The 900 nt gadd7 cDNA does not cross-hybridize to RNA from human or mouse, and so is apparently poorly conserved (35 ). The murine His1 locus is a common site of retroviral insertions leading to murine myeloid leukemia. His1 is conserved across several species, yet its 3 kb transcript has no ORF (36 ). A gene proposed to have a role in differentiation or sex determination (in acampomelic campomelic dysplasia) produces an untranslated 3.7 kb transcript (37 ). DGCR5 is an alternatively spliced, untranslated transcript disrupted in an individual with DiGeorge syndrome (17 ). Finally, the introns but not the exons of the human U22 host gene (UHG) are highly conserved with the mouse homolog, consistent with the hypothesis that the function of the untranslated gene UHG is to provide a host gene for UHG intron-encoded small nucleolar RNAs (38 ).
Both IPW and Ipw are members of a family of untranslated RNA molecules. Although the possibility that the short ORF noted in the Ipw cDNA codes for a polypeptide has not formally been excluded, it is made less likely by the dissimilarity to the ORF in human IPW. Identification of IPW homologs in other organisms may help to identify regions of critical importance to gene function in this poorly conserved gene.
There are two possible roles for untranslated RNAs. First, each RNA product could have a separate cellular function, unrelated to its genomic location, perhaps as part of a ribonucleoprotein complex. The products of the H19, gadd7 and His1 genes all appear to be involved in regulation of cell growth (35 ,36 ,39 ). Expression of H19 and gadd7 inhibits cell growth, and His1 is involved in murine leukemia when activated by retroviral insertion.
Second, genes encoding untranslated RNAs may be important regulators of regional gene expression. In favor of this idea, targeted disruptions of either H19 or Xist in mice cause alterations in the expression of other genes in the cis configuration. Inherited microdeletions in PWS patients that include portions of alternative transcripts of the SNRPN gene causes loss of expression of paternally expressed genes when inherited in the cis configuration through the paternal germline (18 ). Transcription of these genes is thus implicated in control of regional gene expression. Similarly, Ipw could be a candidate for regulation of other genes in the imprinted region of central mouse chromosome 7. Gene knockout studies will be useful to address this possibility. Alternatively, transcripts from an imprinted region in which the chromatin conformation differs on the paternal and the maternal chromosome could simply reflect the opportunistic transcription from a region with open chromatin. Whether such transcription is the prerequisite or the consequence of the underlying mechanism that maintains parent-specific chromatin configuration becomes the crucial unanswered question.
Two phage clones [[lambda]mg1 (13.3 kb), [lambda]mg2 (16.2 kb)] were isolated from a [lambda]FIX II mouse 129 library by hybridization to a probe from IPW exons 2 and 3 [PCR product C-D, (1 )]. Two additional clones [[lambda]67 (10.4 kb) and [lambda]68 (9.5 kb)], were isolated from a 129SvJ embryonic stem cell library by hybridization to a subclone of [lambda]mg1 that included exon C and part of intron C. Three P1 clones [P1-5150 (60 kb), P1-5151 (55 kb), P1-5152 (60 kb)] were identified by PCR screening of a P1 library (Genome Systems) using primers derived from the exon C-intron C. P1 clone DNA was prepared by alkaline lysis. Chromosomal mapping of the Ipw gene by mouse * rodent somatic cell hybrid analysis and fluorescence in situ hybridization was performed as described (21 ).
Both 5' and 3' RACE were performed on mouse brain RNA essentially as described (23 ), with the exception that 5' reverse transcription products were tailed with dGTP rather than dATP. RACE products were cloned into pBluescript (Stratagene) and sequenced. In 5' RACE experiments, reverse primers in exon C were used to isolate clone R5-14 and identify exon B, then reverse primers from exon B were used generate clone R5-49. Forward primers from exon B were used in 3' RACE to isolate clones R3-3, R3-4 and R3-6. Exon B and two copies of exon A were also identified during sequencing of [lambda]mg1. The RACE clones illustrated represent a subset of those identified. Poly(A) tails were found at the 3' ends of clones R3-3, R3-4 and R3-6, indicating that multiple polyadenylation sites are used. Since 3' RACE was performed using an oligo(dT) adapter, we assume that these correspond to 3' ends of Ipw. Clone R3-82-5 was generated by RT-PCR with primers from exons C and F.
A cDNA probe containing parts of exons B and C was used to screen a mouse brain cDNA library (Stratagene #936309), yielding clone mbr0. This clone was used to rescreen the library, yielding 11 additional clones, ranging in size from 1.1 to 5 kb. Clone names and sizes were: mbr0, 652 bp; mbr1, 1.6 kb; mbr2, 1.1 kb; mbr6, 2.2 kb; mbr7, 5 kb; mbr8, 3 kb; mbr9, 1.6 kb; mbr11, 2.9 kb; mbr12, 1.3 kb; mbr15, 2 kb; mbr17, 1.6 kb; and mbr18, 2.3 kb. cDNA library screening was performed by filter lift hybridization, plaque purification and in vivo excision according to the instructions of the manufacturer.
Sequence analysis was performed using the BLAST search algorithm (40 ) through the NCBI WWW site (http://www.ncbi.nlm.nih.gov/BLAST/) and the program CODONPREFERENCE from the GCG package (Genetics Computer Group, Madison WI). Sequencing was performed either using a double stranded dideoxy chain termination method and Sequenase II (USB-Amersham) or by cycle sequencing (Perkin-Elmer Corp.) and an ABI 377 automated DNA sequencer.
Total RNA was prepared from mouse tissues by first grinding under liquid nitrogen, then adding RNA-STAT (TelTestB Inc., Friendswood, TX) and proceeding according to the manufacturer's directions. DNA was extracted from tissues by grinding under liquid nitrogen, proteinase K-SDS treatment and phenol-chloroform extraction followed by ethanol precipitation.
Reverse transcription-PCR was performed using Superscript reverse transcriptase (Gibco-BRL). For expression and imprinting analysis, a hemi-nested PCR strategy was used. RNA was reverse transcribed using random hexamers. The first round of PCR (95oC for 5' then 20 cycles of 95oC for 30 s, 55oC for 30 s, 72oC for 30 s, then 72oC for 5 min) was performed with primers UF2092-UF2077. The second round of PCR, under the same conditions but for 30 cycles, was performed using primers UF2100-UF2077. Amplification of exons B and C from DNA was performed using primers UF2164-UF2165 and UF2135- UF2034 respectively, under the same PCR conditions, for 30 cycles. Primer sequences for imprinting analysis were: UF2092, 5'ACTGTTGTGGTGTTGCTGAC; UF2100, 5'GATGCATTCCTTTTCCTTCA; UF2077 5'TGGTAGAAGAAATGGCACC; UF2164 5'CATCTCAATGTGCATGGTG; UF2165, 5'CTTTGGTCCACAAAGTCTTG; UF2135, 5'CTTTCCCTATGTAGTGTAACCA; and UF2034, 5'AGTTTGTTTAATCATGCATTTT. Oligonucleotides for hybridization and RT-PCR were: exon A, UF2092 (forward); exon B, UF2077 (forward); exon C, UF2100 (reverse); exon D, UF2183 (reverse), 5'ATCCTCCAGGTGGCTTCT; exon E, UF2181 (reverse), 5'GTGCACTGTGCAATTTCTTT; exon F, UF2182 (reverse), 5'AGGGCACCCTTTTTTGATA; exon G, UF2361 (reverse), 5'CCTTCCAAACTGGTTCAGAG; and exon H, UF2362 (reverse), 5'ATAGTTAGGTTTTATGGCCTTC.
We thank Xu Li for the FISH data, David Kingsley for the interspecific hybrid mouse and other mouse DNA samples, Anne Wang and Colin Stewart for genomic clones and helpful discussions and Julie Kerns and Davesh Sharma for technical assistance. R.W. was the recipient of a postdoctoral fellowship from the Medical Research Council of Canada and U.F. is an investigator of the Howard Hughes Medical Institute.
+Present address: Department of Medical Genetics, University of Alberta, Edmonton, AB, Canada, T6G 2S2
Human Molecular Genetics Pages
©
Introduction
Results
Mouse genomic clones hybridizing to IPW map to a conserved syntenic region and contain a partially homologous gene
Characterization of Ipw transcripts and genomic structure
Sequence analysis reveals no coding potential
The Ipw transcript is highly expressed in brain
Ipw is expressed only from the paternally derived allele
Discussion
Materials And Methods
Identification of genomic clones
RACE and identification of cDNA clones
Sequencing and sequence analysis
RNA and DNA preparation
PCR and RT-PCR
Acknowledgements
References
REFERENCES
This page is maintained by OUP admin. Last updated Fri Jan 31 11:45:11 GMT 1997. Part of the OUP Journals World Wide Web service.Copyright Oxford University Press, 1996